HAL Id: hal-02470909https://hal-univ-rennes1.archives-ouvertes.fr/hal-02470909

Submitted on 17 Feb 2020

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

The influence of host-plant connectivity on fungalassemblages in the root microbiota of Brachypodium

pinnatumCendrine Mony, Nathan Vannier, Philomène Brunellière, Marine Biget, Sophie

Coudouel, Philippe Vandenkoornhuyse

To cite this version:Cendrine Mony, Nathan Vannier, Philomène Brunellière, Marine Biget, Sophie Coudouel, et al.. Theinfluence of host-plant connectivity on fungal assemblages in the root microbiota of Brachypodiumpinnatum. Ecology, Ecological Society of America, 2020, 101 (4), pp.e02976. �10.1002/ecy.2976�.�hal-02470909�

Article type : Articles

5

The influence of host-plant connectivity on fungal assemblages in the root microbiota of

Brachypodium pinnatum

Authors

10 Mony C1*, Vannier N1,2, Brunellière P1, Biget M. 1, Coudouel S. 1, Vandenkoornhuyse P.1

Addresses

1UMR 6553 Ecobio, CNRS - University of Rennes, Avenue du Général Leclerc, 35042

Rennes Cedex, France ;

15 2Current address: Department of Plant Microbe Interactions, Max Planck Institute for Plant

Breeding Research, 50829 Cologne, Germany.

*: corresponding author, Cendrine.mony@univ-rennes1.fr

Running title:

20 Host-plant connectivity and microbiota

Acc

epte

d A

rtic

le

Abstract

Dispersal limitation may drive the structure of fungal microbiota of plant roots at small spatial

25 scales. Fungal root microorganisms disperse through the plant rooting systems from hosts to

hosts. Due to a pronounced host-preference effect, the composition of endophytic root

microbiota may follow plant distribution. A given plant community may hence include a

matrix of host-plant species that represent various habitat permeabilities to fungal dispersal in

the floristic landscape. We experimentally tested the effect of host-plant isolation on

30 endophytic fungal assemblages (Ascomycota, Basidiomycota, Glomeromycotina) inhabiting

Brachypodium pinnatum roots. We calculated host-plant isolation using Euclidean distance

(distance-based dispersal limitation) and Resistance distance (functional-based dispersal

limitation), based on host presences. All fungal groups were more influenced by the resistance

distance between B. pinnatum than by the Euclidean distance. Fungal dispersal was hence

35 strongly related to the spatial distribution of the host-plants. The fungal groups displayed

however different responses (in richness, abundance, and composition) to host isolation.

Additionally, fungal assemblages were more strongly controlled by the degree of connectivity

between host-plants during the prior year than by current connectivity. This discrepancy may

be due to changes in plant species coverage in a year and/or to the delay of dispersal response

40 of fungi. This study it the first to demonstrate how small-scale host-plant distributions

mediate connectivity in microorganisms. The consequences of plant distributions for the

permeability of the floristic landscape to fungi dispersal appear to control fungal assemblages,

but with possibly different mechanisms for the different fungal groups.

45 Keywords

Plant-fungi interaction, host preference, fungal dispersion, assembly rules, root fungal

microbiota, micro-landscape

Introduction

50 Understanding the assembly rules of plant microbiota is a central question of plant biology

because of the tremendous ecological functions provided by symbiotic microorganisms to

plants (Davison et al. 2016, Lekberg and Waller 2016). Next generation sequencing

approaches have demonstrated that the microbiota of plants is composed of a large diversity

of microorganisms, including bacteria (Lundberg et al. 2012, Bulgarelli et al. 2012), archaea

Acc

epte

d A

rtic

le

55 (Edwards et al. 2015, Vannier et al. 2018) and fungi (Vandenkoornhuyse et al. 2002,

Coleman-Derr et al. 2016, Lê Van et al. 2017). An increasing number of studies suggests that

the identity of the host-plant species (Vandenkoornhuyse et al. 2002, Martinez-Garcia and

Pugnaire 2011, Yang et al. 2012) and to some extent, its genotype (Bodenhausen et al. 2014,

Edwards et al. 2015) condition the associated microbiota through the preferential recruitment

60 of particular fungi. This host-preference effect has been demonstrated in Arbuscular

Mycorrhizal Fungi (AMF) (Vandenkoornhuyse et al. 2002). It was interpreted as the

consequence of a plant filtering and a carbon reward process to the AMF best cooperators

(i.e., providing the highest amounts of phosphorus to the host plant in exchange for carbon)

(Kiers et al. 2011, Duhamel and Vandenkoornhuyse 2013). Such host-preference has been

65 overlooked for the other components of the endophyte microbiota. If this host-plant

preference exists, the plant community at any given location should be regarded as a mosaic

of habitats with some plant species providing more favorable niches than others for any given

fungal symbiont. Plant species distributions are then likely to affect the assemblages harbored

among the roots of individual plants, explaining the observed heterogeneous distribution of

70 fungi.

Island biogeography theory explains how habitat isolation affects species diversity

(MacArthur and Wilson 1967). This theory assumes that in fragmented landscapes, local

populations are linked through the dispersal fluxes of individuals and can hence be structured

as a network (metapopulation). This theory has been extended to metacommunities (Wilson

75 1992, Leibold et al. 2004). This conceptual framework assumes that there are four main

drivers of local community dynamics: patch disturbances, local abiotic and biotic

environmental conditions on species sorting, dispersal among communities, and stochastic

extinction and colonization events (Leibold et al. 2004, Chase 2005). When dispersal is the

main driver (i.e., dispersal rates are high), the species composition of local communities tends

80 to become more homogeneous. Species richness at the metacommunity scale is also expected

to be higher because of the lower probability of genetic drift, which counteracts the higher

probability of stochastic extinctions observed in isolated populations (Hanski 1999). Isolation

between local communities was at first approximated by a geographic distance. Isolation by

distance has been transposed and tested recently on microorganisms (i.e. distance-decay

85 relationships, Bell, 2010; Hanson et al. 2012). But in many species dispersal is not only linked

with the geographical distance but also driven by landscape connectivity (i.e. the degree to

which a given landscape facilitates or impedes the movement of organisms among habitat Acc

epte

d A

rtic

le

resources) (Taylor et al. 1993). Connectivity is then dependent on the way landscape elements

are distributed over space (structural connectivity) and on how species disperse over these

90 elements (functional connectivity) (Tischendorf & Fahrig, 2000; Calabrese & Fagan, 2004;

Taylor et al., 2006). High connectivity increases biodiversity for a large set of

macroorganisms (Fletcher et al. 2016). The application of this framework linked to

biogeography and landscape ecology to microbial assemblages associated to plant hosts has

been mostly developed toward the understanding and the prediction of specific species spread

95 such as pathogens (Thrall et al. 2003, Sullivan et al. 2011). While considering the whole

fungal assemblages, the use of island biogeography theory for understanding the fungal

microbiota associated with plants has mostly focused on ectomycorrhizal fungi (Peay et al.

2007, Peay et al. 2010) and at large landscape scales, similar to the scales used to study

dispersal among macro-organisms.

100 Recent findings advocated however that fungal dispersal patterns should be examined

at small spatial scales because of the high heterogeneity detected at such scales for some

groups of fungi (such as AMF) (Bahram et al. 2015). The dispersal of fungi is achieved by

dispersing spores and/or by expansion of the hyphal network (Carvalho et al. 2004). However,

the relative contribution of these two dispersal types for plant root colonization is difficult to

105 determine (Smith and Read 2008). In some mycorrhizal species, large spore size and

belowground spore production restrict propagation distances (Smith and Read 2008).

Likewise, hyphal growth is limited in space (Hart and Reader 2002), leading to the

assumption that dispersal-limitation likely occurs at small scales for many fungal species. For

biotrophic species, fungal species must disperse from one host plant to another. A plant

110 community representing a matrix of different plant species constitutes a variety of potential

fungal habitats (floristic landscape) through which fungus species must disperse. It can be

imagined that the ability of a fungus species to disperse among plants depends on its host

preferences and its dispersal traits, as well as on the root attributes of its host-plants. For a

given host-plant in a given floristic landscape, connectivity with individuals of the same

115 species is then likely to increase the rate of fungi dispersal between these host plants.

This study aims to test the effect of host isolation on the composition of root fungal

assemblages by focusing on endophytic fungi assemblages of Brachypodium pinnatum. We

used a mesocosm experiment presenting a large range of B. pinnatum spatial configurations in

order to test the effect of B. pinnatum connectivity degree on the fungal composition

120 associated to roots of the host-plant B. pinnatum. We calculated the degree of isolation of the Acc

epte

d A

rtic

le

sampled host-plants with Euclidean distance (i.e. distance-based dispersal limitation) and

resistance distance (i.e. functional dispersal limitation (sensu Fagan and Calabrese 2006)). We

expected that more connected host-plant species within a floristic landscape result in a higher

fungal species richness at the metacommunity scale (i.e. group of host-plants), and at the

125 community scale (i.e. individual host-plant) and in closer fungal microbiota compositions

between hosts. In contrast, isolated host-plant species within a floristic landscape should

result in a lower fungal species richness at the metacommunity and community scales and

should harbor different sets of fungal microbiota.

130 Methods

Sampling host-plants

We investigated the composition of plant microbiota associated with the focal plant B.

pinnatum. This Poaceae species is a widely-distributed species of temperate grasslands. It

displays a clonal plant that forms aggregated patches thanks to its phalanx growth form (sensu

135 Lovett-Doust 1981). B. pinnatum root-associated mycobiota has been shown to differ

depending on the surrounding plant composition and configuration (Bittebiere et al., 2019;

Mony et al., 2019). Individual B. pinnatum were sampled in 16 outdoor mesocosms (1.30 x

1.30 x 0.25 m) encompassing four different compositions of herbaceous grassland plant

species, following an additive design, and with four replicates for each treatment: (1) a B.

140 pinnatum monoculture, (2) a two-species mixtures comprised B. pinnatum and Elytrigia

repens, (3) a four-species mixture comprised B. pinnatum, Agrostis tenuis, Festuca rubra, and

Anthemis vulgaris, and (4) an eight-species mixture comprised of the four species of mixture

#3, plus Elytrigia repens, Agrostis stolonifera, Holcus mollis, and Ranunculus repens. These

nine herbaceous species are widely distributed in grasslands on western France (des Abbayes

145 et al. 1971).

The mesocosms were established in 2009 in outdoor experimental gardens

(regular watering but no control of light, rainfall, insect herbivory) three years before we

sampled the fungi associated with B. pinnatum roots. We assumed that this 3 year duration

after establishment of the mesocosm guarantees the fact that the host-preference effect is the

150 main driver of host-plant recruitment of fungi over priority effect. Each mesocosm treatment

consisted of a mixture of 48 plantlets grown in a hexagonal pattern. The density of the 48

plants was equally distributed among species in the mixtures (e.g. for the four-species

Acc

epte

d A

rtic

le

mixtures, 12 plantlets of each per species were planted in each mesocosm). Plantlets were

randomly assigned to each position within the hexagonal pattern (Appendix S1: Fig. S1).

155 Inflorescences were periodically cut to suppress seed production and all new seedlings were

removed manually. This ensured that the plant cover is the result of the initially planted

individuals that developed through clonal multiplication without any individuals coming from

other mesocosms. In absence of existing knowledge, we assumed that the stress induced to the

plant by the cut has a minor effect on fungi composition and, if it exists, this potential effect is

160 similar across sampled plots. Mesocosms were mown each year in fall and biomass was

removed. Mesocosms were watered during the dry season to avoid drought stress. Plants

colonized bare soil in contrasting distribution patterns, even within the same type of mixture

(treatment) due to random clonal propagation of individuals. Therefore, the 16 mesocosms

(plots) varied widely in their spatial distribution of B. pinnatum, meaning that plots

165 encompassed a wide range of B. pinnatum isolation intensities (Appendix S1: Fig. S2).

Mesocosms varied widely in plant species compositions and richness, allowing us to test our

expectations across a large range of plant communities. Within our experimental plots, five B.

pinnatum individuals, one located near each of the four corners of the plot at a minimum

distance of 20cm to the border, and one at the center of the plot (Appendix S1: Fig. S2) were

170 sampled. Due to the patchiness of B. pinnatum in some plots, three individuals in four of the

plots and four individuals in two of the plots were sampled.

Measuring current and past connectivity between host-plants

Two metrics were used to assess the connectivity between pairs of focal plants: Euclidean

175 distance and resistance distance (Appendix S1: Fig. S3). Euclidean distance corresponds to

the degree of isolation (by geographic distance alone) between pairs of plants, which models

degree of connectivity due to distance-isolation. Euclidean distance is independent of the

presence of the focal plants (B. pinnatum) or other plant species. Resistance distance

corresponds to the degree of isolation due to landscape resistance to dispersal. It is based on

180 circuit theory (McRae et al. 2008). It assumes that a landscape consists of a mosaic of patches,

each with a different resistance to dispersal. Less resistant areas are assumed more permeable

to dispersal, whereas areas that are more resistant inhibit dispersal (less permeable). In the

present study, we assumed that the dispersal of fungi associated with B. pinnatum is

preferentially achieved along paths inhabited by other B. pinnatum plants than along paths of Acc

epte

d A

rtic

le

185 other species. This distance is considered as a functional-based connectivity index as it takes

into account the functional response of the microbiota to the spatial distribution of host-plants.

To calculate Euclidean and resistance distances, B. pinnatum sampled host plants were

assimilated as a graph where the sampled plants represented nodes. The distances for each

pair of nodes were calculated using the software ArcGis for the Euclidean distances and the

190 software Circuitscape for the resistance distances (McRae et al. 2008). The resistance distance

for a given pair of nodes was calculated based on the cumulated resistance for all possible

paths between the two nodes (plants) relatively to the probability that the path would be used

for dispersal. Resistance distances were defined based on B. pinnatum distribution maps

because we defined low resistant areas as patches where B. pinnatum occurred and high

195 resistant areas as areas where the host-plant was absent. Resistance distances were calculated

for each pair of nodes based on the presence of B. pinnatum in an 80 x 80-cm grid of 256

cells, each cell being of 5 x 5 cm in size. This grid was positioned at the center of each plot. A

plant was considered to be present in a grid-cell if at least one shoot was rooted within it. To

take into account the plant-plant competition and resulting plant distribution changes, plant

200 communities were mapped in 2011, one year before we began sampling them (i.e., the ‘past’

map) and again in 2012 (i.e., the ‘current’ map), the same year that we sampled B. pinnatum

individuals. Assessing resistance distances for each of these dates is necessary for taking care

of any potential lag-time response of microbiota to host plant isolation,

Euclidean and resistance distances were then computed for both maps. Because B.

205 pinnatum has a very superficial and narrow rooting system, we assumed that the grid cells

adequately reflected root distributions at a 5-cm spatial resolution. Each grid-cell of the

resistance maps represented a cost for dispersal (1 = low cost), which depended on the

presence of B. pinnatum in the cell. Patches where B. pinnatum occurs had a low cost of 1.

When B. pinnatum was absent in a given grid cell, the cell was either occupied by another

210 plant species or by bare soil (infrequent). Where B. pinnatum was absent, we determined a

specific cost of dispersal, which depended on the relative effort required for a fungal species

to disperse through such a grid cell relative to cells that harbored B. pinnatum. In order to test

the sensitivity of the index to the cost attributed to the absence data (cells without B.

pinnatum), we tested several relative dispersal costs: 10, 30, 50, 70, and 100. For example, a

215 dispersal cost of 10 for a cell meant that in the grid-cell, fungi might be 10 times less likely to

disperse through it than through a grid-cell occupied by B. pinnatum. Resistance distances

were then calculated for each pair of the sampled host-plants. We determined the correlation Acc

epte

d A

rtic

le

between resistance distances calculated with the various magnitudes of costs we tested for

grid-cells without B. pinnatum. Pearson coefficients between resistance distances obtained

220 with costs of 10, 30, 50, 70 and 100 attributed to cells without B. pinnatum were all >0.99

(p<0.001) both for 2011 and for 2012. This suggests that dispersal paths simulated were

independent on the ratio of dispersal costs in absence or presence of B. pinnatum in the range

10-100. Therefore, the following dispersal costs were assigned to cells: 1 (for cells with B.

pinnatum) and 50 (to cells without B. pinnatum).

225 We calculated two metrics to represent the degree of isolation for sampled host-plants

based on resistance distances. One metric was the mean of pairwise resistance distances for a

given host-plant with all other host-plants (4 pairs) and the other was the mean of the pairwise

resistance distances relative to the nearest host-plants, with the number of pairs examined

dependent on the location of the sampled plants in the plot (i.e. based on three pairs if the

230 sampled plant was located in a plot corner and based on four pairs if the plant was located at

the center of the plot). Because both metrics were highly correlated with one another (Pearson

coefficient, R=0.99; p<0.001 for 2012, R=0.98; p<0.001 for 2011), we presented only the

results for the mean pairwise resistance distances based on the nearest host-plants. Hereafter,

these pairwise comparisons represent the “sample scale.” The degree of isolation of B.

235 pinnatum plants was also calculated at the plot scale. This was determined by the mean

resistance distance between all pairs of sampled individuals (10 pairs). This characterized the

average connectivity of the plant community (hereafter, these means represent the “plot

scale”). The degree of isolation at the sample and plot scales was not calculated based on

Euclidean distance because all samples and plots displayed the same values (all samples

240 nearly equidistant in the plot).

Assessing root fungal assemblages

We collected ~ half of the rooting system of a B. pinnatum individual plant in each of the

sampled mesocosms. Roots of each sample were cleaned in a 1‰ solution of Triton X100,

245 and thoroughly rinsed with tap water then sterile distilled water. For a given sampled plant,

~80 mg out of at least ~1g of fresh mass were subsampled. Healthy root fragments were

selected for being representative of the different parts of the rooting system. After this

cleaning step, only root-adhering epiphytic and endophytic microorganisms were still present

with the roots (e.g. Lê Van et al. 2017). Cleaned roots of each sample were ground to powder Acc

epte

d A

rtic

le

250 under liquid nitrogen using a pestle and mortar. Total DNA was extracted from the 80mg sub-

sample using the DNeasy plant kit (Qiagen) according to the manufacturer’s

recommendations. We amplified a specific 550 bp of the fungal SSU rRNA gene fragment,

including the V4 and V5 regions, using the primers SSU0817 and -NS22B following

procedure outlined by Borneman and Hartin (2000) and Lê Van et al. (2017). The two primers

255 were modified on their 5’ end with the addition of Illumina® tails. The PCRs performed with

Illustra™ PuReTaq Ready-to-go (GE Healhcare®) contained 0.2 μM of primer within a final

volume of 25 μL following the procedure of Lê Van et al. (2017). This first PCR was

followed by a second PCR to tag the amplicons. Amplicons were purified in Agencourt®

(AMPure™) solution using a Bravo Automated Liquid Handling Platform (Agilent®).

260 Purified amplicons were then quantified with a Quant-iT™ PicoGreen™ dsDNA Assay Kit

and normalized to the same concentrations before a second PCR to conduct multiplexing

tagging, which was performed with a Smartchip Wafergen/Takara instrument. The pooled

PCR products were purified and quantified by qPCR before pair-end sequencing (Illumina®,

Miseq). Both the sequencing library and mass sequencing were performed using the Human

265 and Environmental Genomics platform (https://geh.univ-rennes1.fr/?[en]).

The FROGS pipeline was used to analyze amplicon sequences (Escudié et al. 2017).

The clustering within FROGS was performed using SWARM (Mahé et al. 2014) a new

generation method allowing to avoid the use of a threshold. To limit the risks of obtaining

artificial OTUs (Operational Taxonomic Units), we only retained identical sequences

270 observed in at least five different samples. OTUs were affiliated to taxa by comparing the data

to the Silva 18S database, release 132 (Quast et al. 2013). We built a contingency matrix from

the data. The number of sequences was normalized at the same size for all the samples by

rarefying to the smallest number of sequences observed (i.e. 19,224 sequences for each

sample) using the VEGAN package in R (function ‘rarefy’; Oksanen et al., 2018; Rarefaction

275 plot provided in Appendix S1: Fig. S4).

From this normalized matrix, richness and evenness of OTUs were calculated for the

whole fungi assemblage we obtained from the roots of the sampled plants collected from each

plot. We additionally analyzed separately the different fungal groups because the effects

detected at the scale of the entire fungal community might be driven by the most dominant

280 phylum. This allows also to focus on interesting taxonomic groups such as the sub-phylum

Glomeromycotina which was previously proven of major importance for plant growth (Van

der Heijden et al., 1998). Richness at the plot scale was calculated as the total number of

OTUs for all sampled host plants in a plot, and at the sample scale as the number of OTUs

Acc

epte

d A

rtic

le

that were associated with roots of the sampled plant. Evenness was calculated only at the

285 sample scale using Piélou evenness index Dissimilarity indices were also calculated for each

B. pinnatum pair based on Bray-Curtis dissimilarity index using the VEGAN package

(function ‘vegdist’). This dissimilarity index for fungal assemblages was used as a proxy to

determine the extent of dispersal between pairs of B. pinnatum. They are then interpreted as

follows: the higher the dissimilarity index, the more dissimilar the assemblages are in

290 composition.

Statistical analyses

Dissimilarity in fungal community composition was described through a PCoA on bray-curtis

distances between samples computed on the normalized OTUs table for both the plot scale

295 (Appendix S1: Fig. S5) and the sample scale (Appendix S1: Fig. S6) (package “vegan” and

“phyloseq”; McMurdie & Holmes, 2013). We used then a linear model to analyze the effect

of host-plant isolation in 2011 and in 2012 on fungal OTU richness at the plot scale. Host-

plant isolation corresponded to the mean of pairwise resistance distances over all possible

pairs of host plants in a given plot. To analyze the effect of a given host-plant’s isolation on

300 the richness of its fungi assemblage and its evenness at the sample scale, we applied linear

mixed models using the mean pairwise resistance distance for a given host plant to its nearest

sampled host plants (resistance distance for 2011 or 2012) as the independent variable and

fungal species richness (or evenness) as the dependent variable. OTUs richness was log-

transformed to ensure normal distributions. Plots were integrated in the model as a random

305 factor. We did not test for the effect of Euclidean distance on OTU richness and evenness

because all samples had similar distance to the nearest neighbor (sample scale) nor mean

distance between sampled plants (plot scale) due to the standardized sampling design

(sampling in the four corners and in the center). The effect of Euclidean distance was tested

though on the similarity index

310 The effect of host-plant isolation on fungi similarity between pairs of sampled

individuals were analyzed using linear mixed models. The independent variable was isolation

distance for pairs of sampled individuals with three different distances tested (Euclidean

distance, Resistance distance calculated for 2011 and Resistance distance calculated for

2012). The dependent variable was the dissimilarity between the fungi assemblages measured

315 using the Bray-Curtis index for a total of 3 models per fungal group. In these models, the

sampled individuals of each pair were considered as random factors nested into the plot factor

Acc

epte

d A

rtic

le

(i.e. three random factors with two nested into one). This structure enabled us to manage the

partial dependencies in the data.

The independence and normality of the residuals for all tested models were examined

320 graphically. The significance of each model was tested using an ANOVA based on the Type

II sums of squares and calculated the proportion of variance explained by the fixed effect

(marginal R²). The proportion of variance explained by random effect(s) (conditional R²)

(sensu Nakagawa and Schielzeth 2013) was calculated for mixed models. We used the

following software packages for statistical analyses: ‘nmle’ (Pinheiro et al., 2018), ‘lme4’

325 (Bates et al. 2015), ‘MuMIn’ (Barton 2016), and ‘car’ (Fox & Weisberg 2011) in R.3.0.4 (R

Core Team).

Results

Fungi assemblage characteristics

330 B. pinnatum roots were colonized by 1006 fungal OTUs described by 1,345,680 sequences

(Fig. 1). The root fungal microbiota of the entire fungi assemblage were dominated by

Ascomycota (67.1% of the sequences and 61.8% of OTU richness). Basidiomycota

constituted 25.9% of the sequences and 25.0% of OTUs, whereas Glomeromycotina

constituted 5.1% of the sequences and 5.4% OTU richness. The dominant classes of the

335 Ascomycota were Dothideomycetes (78.0% of Ascomycota sequences and 22.5% of OTU

richness; orders: Jahnulales, Pleosporales). Sordariomycetes constituted 12.5% of

Ascomycota sequences and 35.6% of OTU richness (orders: Boliniales, Hypocreales,

Magnaporthales, Microascales, Ophiostomatales, Sordariales, Xylariales), whereas

Leotiomycetes constituted 8.7% of Ascomycota sequences and 20.4% of OTU richness

340 (orders: Erysiphales, Helotiales, Medeolariales, Thelebolales). Other classes were lower than

1% of total sequences. The dominant class of the Basidiomycota was the Agaricomycetes

(99.0% of Basidiomycota sequences and 90.1% of OTU richness; orders: Agaricales,

Geastrales, Trechisporales, Gomphales). The dominant class of the Glomeromycotina was

Glomerales, which represented 99.99% of the sequences and 87.0% of OTU richness. The

345 composition in fungal OTUs varied among assemblages both at the plot (Appendix S1: Fig.

S5) and sample scales (Appendix S1: Fig. S6). Acc

epte

d A

rtic

le

Effect of host isolation on fungal assemblages assessed at sample and plot scales

Indices used to characterize the fungal microbiota in B. pinnatum roots are provided in Tables

350 1 & 2. OTU richness and evenness for “all fungi”, Asco-, Basidio-, and Glomeromycotina

varied widely between samples, both at sample and at plot scales. We did not test the effect of

Euclidean distance on fungal richness and evenness as all samples and plots displayed similar

values of isolation due to the sampling design (see method section). This means that the

strong variability of the OTUs richness and evenness observed is not related to changes in

355 geographic distance. The results testing the effect of the resistance distance on fungal richness

and evenness are presented below.

At the plot scale (Table 1), only one statistical model of the eight we produced

was significant (R² = 0.20; p=0.047). For “all fungi”, Ascomycota and Glomeromycotina, we

detected no significant relationship between degree of isolation of B. pinnatum and OTU

360 richness of fungal microbiota inhabiting its roots. For Basidiomycota, we detected a

significant effect of host isolation on OTU richness only for 2011. Host isolation decreased

Basidiomycota richness, as indicated by the negative Estimate (‘Est’) value of -0.1.

At the sample scale (Table 2), five of 16 models were statistically significant with

marginal R² ranging from 0.06 to 0.13. In Ascomycota and Basidiomycota, we did not detect

365 a significant effect of OTU richness on resistance distance in 2011 or 2012. In contrast,

Glomeromycotina OTU richness increased with mean resistance distance in both the 2011 and

2012 samples, as indicated by the positive Estimate (‘Est’) values of 0.009 and 0.02. We

found no effect of mean resistance distance of 2012 on evenness for Asco-, Basiodio-, or

Glomeromycotina, nor was mean resistance distance of 2011 related to evenness in

370 Ascomycota. The effect of mean resistance distance of 2011 on evenness differed between

fungi assemblages: higher values of evenness for “all fungi” and Basiodiomycota assemblages

occurred in more isolated B. pinnatum plants (positive ‘Est’ of 0.001 and 0.003 respectively in

Table 2), whereas low values of evenness for Glomeromycotina assemblages occurred in

more isolated B. pinnatum plants (negative ‘Est’ of -0.004 in Table 2).

375

Effect of host-isolation on the similarity of fungal assemblages between B. pinnatum plants

Euclidean distance between sampled B. pinnatum plants was not correlated with resistance

distance (Pearson coefficients: R= 0.04; p>0.05 for resistance distance in 2011; R= -0.07;

p>0.05). Resistance distance in 2012 was significantly correlated with resistance distance in

380 2011, although not very strongly (R = 0.43; p<0.001). Acc

epte

d A

rtic

le

For “All fungi” and Ascomycota colonizing B. pinnatum roots, the Bray-Curtis

dissimilarity index declined with degree of host-plant isolation with Euclidean and resistance

distance in 2011 (Table 2). Assemblages were more similar when host plants were isolated

than when they were more connected to other B. pinnatum (i.e. lower resistance distance)

385 (negative ‘Est’ of -0.002 and -0.003 respectively). We did not detect a significant relationship

between the Bray-Curtis index and Euclidean distance or resistance distances for

Basidiomycota nor Glomeromycotina.

Discussion

390 We demonstrated the effect of B. pinnatum isolation on fungal microbiota assemblages. This

effect is robust for the relationships between isolation and the richness and evenness of the

assemblages both at the plot and sample scales. For dissimilarity indices however,

significances of signals were low. This can be due to the limitation of the Bray-Curtis indices

to fully and quantitatively represent dispersal fluxes of fungi species between local

395 communities. These indices can also reflect possible other mechanisms than dispersal such as

environmental similarities. Despite homogeneous environmental conditions of soil, water or

light of the mesocosms, we cannot exclude that there is no heterogeneity at very fine grain,

possibly due to the effect of plants themselves over the duration of experiment. Slight nutrient

or water depletion in the soil below more competitive plant individuals is likely to occur, as

400 well as slight increase in herbivore pressure around more palatable plant individuals. This

fine-scale heterogeneity reflects real conditions of grassland communities despite their

possible environmental homogeneity at the field scale. Discussion is made below within these

limitations.

405 Host-plant connectivity affects plant fungal microbiota assemblages

As expected, we demonstrated the effect of the degree of host-plant isolation on OTUs

composition, richness, and evenness (Tables 1, 2 & 3). Differences in plant fungal microbiota

assemblages depended on the degree of host-plant (B. pinnatum) isolation, which varied

widely and were independent of plant species richness and plant composition. Interestingly,

410 the effect of the degree of plant isolation on plant fungal microbiota differed among fungal

groups.Acc

epte

d A

rtic

le

For “all fungi” and Ascomycota, OTU assemblages between host plants were more

similar in isolated populations as measured by resistance distance, whereas they differed in

relative abundance among highly-connected populations (Table 3). This suggests that when

415 there is no limitation to dispersal, OTUs relative abundances in fungal assemblages of the

focal plant species (B. pinnatum in this study) are essentially regulated by other parameters

acting at local scale. We can speculate that they are due to slight changes in environmental

conditions (abiotic or biotic) or fungal traits that should drive assemblages in the local patches

(species sorting paradigm, sensu Leibold et al. 2004).

420 In Basidiomycota, the degree of isolation as measured as resistance distance, was

related to OTU richness at the plot scale (i.e. the entire assemblage recorded over the sampled

individuals of our plots) (Table 1), suggesting that Basidiomycota are more impacted by the

degree of isolation of their host plants (habitat fragmentation at the “metacommunity” scale)

than Ascomycota or Glomeromycotina are. In these later two groups, we can speculate that

425 their higher resistance may be due to traits that promote efficient dispersal ability between

isolated host plants. For instance, large spores in Glomeromycotina have been shown to be

more effective than small spores in initiating root colonization (Daniels et al. 1981), whereas

rapid hyphal growth and particular hyphal architectures are known to promote root

colonization at small spatial scale (Hart and Reader 2002).

430 Our study showed that the response of Glomeromycotina to host-plant isolation

(measured as OTU richness) is higher and more equitable in roots of more isolated host plants

than in more connected ones (Table 2). They tend also to be more similar while connected

though this result needs to be supported by further experimentations as it is only a statistical

tendency. Isolated host plants are surrounded by other plant species. These neighboring

435 species may display different AMF assemblages and thus enrich the local soil with them, from

which AMFs are recruited by B. pinnatum. Fungal assemblages of neighboring plants have

indeed been shown to impact the AMF root composition of plants located within a few

centimeters from them (Shi et al. 2014; Bittebiere et al. 2019).

440 Effect of host-plant connectivity is not necessarily linked to their geographic distance

“All fungi”, Ascomycota, Basidiomycota, and Glomeromycotina appeared to be more

dependent on resistance distance than on Euclidean distance between host-plants (Tables 1, 2

& 3) indicating that the isolation-by-distance connectivity model is not adequate for

Acc

epte

d A

rtic

le

endophytic fungi. We indeed only recorded for Ascomycota group a tendency to depend on

445 Euclidean distance while no effect was detected for the other groups. The Ascomycota group

includes a morphologically diverse group of species, including single-celled yeasts,

filamentous fungi, and even more complex fungi, thus representing a wide range of dispersal

mechanisms and patterns (Bennett and Turgeon 2017). Additionally, within the endophytic

assemblages, some species are not obligatory biotrophs and so their ability to disperse may be

450 independent of whether a suitable host is present. If we could conduct a similar study based

on more precise taxonomic subdivisions, it might be possible to obtain more insight into

Ascomycota dispersal ecology and their response to the connectivity models.

Our results suggest then that fungal dispersion is more strongly related to the spatial

distribution of host-plants at the plot scale than by geographic distance; supporting existing

455 empirical results. For example, Carvalho et al. (2003) found that spore abundance in soil is

related to the proximity of mycotrophic hosts. Similarly, Shi et al. (2014) demonstrated that

the composition of Glomeromycotina species in the roots of an invasive plant species and in

its neighboring plants depended on the density of the invasive species in the patch and hence

its higher probability of being in closer proximity to other individuals of the same species.

460 When resistance distance is low, individuals of host plants tend to be aggregated, thus

providing a higher probability of plant rooting systems being in contact with one another.

Transfers of fungi among aggregated host plants of a given species means that they are more

likely to harbor more similar fungi microbiota (i.e. via direct contacts between rooting

systems or hyphal growth) (McGee at al. 1997, Smith and Read 2008). Propagules produced

465 by fungi are also expected to form similar symbiotic reservoirs.

Resistance distance rather than geographic distance has proved to be a better predictor

of the effects of patch isolation on macro-organisms (McRae and Beier 2007). In our study,

we were able to use a similar approach to gain insight into dispersal potentials among

microorganisms but we used a simple binary distribution of resistance costs based on the only

470 presence/absence of the host-plant B. pinnatum. Refinements to our approach might be

possible by imposing different permeability degrees (i.e. resistance costs) for each plant

species comprising the plant community. Endophytic fungal assemblages are preferentially

associated with specific plant species but some other plant species might constitute alternative

hosts (Gollotte et al. 2004). Fixing permeability degrees needs to better describe fungal

475 microbiota over a large number of plant species in order to analyze the similarities and

divergence of fungal composition between host species. Studies suggest that AMF community Acc

epte

d A

rtic

le

is determined by the taxonomic family of the plant species (Scheublin et al. 2003, König et al.

2010) or the plant life form (Yang et al. 2012, Lopez-Garcia et al. 2014). These attributes

could be used to predict similarity of microbiota and ultimately calibrate the resistance costs.

480 In addition, permeability is assumed here to be similar for all fungi while the cost of dispersal

is likely to differ between fast- and slow-growing fungi and between those that can use

vectors or have a high specificity and those that are not transported by vectors or have a low

specificity. This work paves the way for studies looking at studies on connectivity effect

based on more refined interactions between fungi and host-plant isolation and extended to the

485 entire endophyte microbiota.

Past B. pinnatum connectivity drives plant fungal microbiota assemblages more than present

connectivity

For “all fungi” and the Asco- Basidio-, and Glomeromycotina groups, characteristics of

490 fungal assemblages were equally or better explained by degree of connectivity in 2011 than

degree of connectivity in 2012 (the year we sampled B. pinnatum microbiota). This

discrepancy may be partly due to changes in plant species coverages that occurred between

2011 and 2012 (Fig. 2; 0.5 to 3.3 fold increase in cover respectively in B. pinnatum

monoculture and the eight-species mixture). This led to an overall decline in the degree of

495 host-plant isolation in plots in 2012 (isolation assessed with B. pinnatum presence/absence

data).

An alternative explanation of the delay of response of the plant fungal assemblages to

changes in vegetation distribution might be due to the time necessary for dispersing in these

microorganisms. Many root fungi produce propagules to disperse, which in some Ascomycota

500 and Basidiomycota may be produced either aboveground and/or in soils. However,

Glomeromycotina only produce propagules belowground, likely leading to a low probability

of long-distance dispersal events. A large proportion of fungi do not sporulate at all or

sporulate infrequently (Clapp et al. 1995). In these types, a large part of dispersal is achieved

by hyphal growth and infections by root contacts between plants (Carvalho et al. 2004). Thus,

505 vegetative multiplication in these types of fungi occurs then within a few centimeters around a

given rooting system (Warner and Mosse 1980) leading to a slow dispersal from hosts to

hosts. Lastly the timing of fungi dispersal may also depend on host plant phenology (Johnson-

Green et al. 1995) and especially on the growth and architecture of host roots (Bundrett and Acc

epte

d A

rtic

le

Kendrick 1990). Additional studies, with controlled experimental designs that manipulate

510 various configurations of host plants may help to deepen our understanding of dispersal

mechanisms among root-associated fungi.

The concept of connectivity for microorganisms: toward a better understanding of fungi

assembly rules

515 The importance of connectivity in regulating dispersal has been a central ecological question

for understanding assembly rules in macro-organisms. Our study is the very first to apply

these approaches to better understand dispersal at a small scales (centimeters) and obtain

insight into how plant microbiota communities assemble. We demonstrated that dispersal

limitation in root fungi was less determined by distance than by the spatial arrangements of

520 host plants and their consequences on the permeability of the plant landscape to dispersion.

Such permeability depends on the presence of specific host plants, thus confirming the

occurrence of a host-preference effect (Vandenkoornhuyse et al. 2002). We found that this

effect not only occurs in Glomeromycotina fungi, but that it also occurs in Ascomycota and

Basidiomycota.

525 In agreement with the common consensus for macroorganisms, low connectivity

between plant hosts is associated with a lower richness of the species pool of Basidiomycota

fungi. However, in Glomeromycotina, localized host isolation promotes the effect of

neighbouring patches, illustrating a possible edge effect, well-known in landscape ecology for

promoting species richness (Ries et al. 2004). Such differences in response to degree of

530 connectivity between fungi groups suggest that edge effect vs. dispersal drivers differ in their

relative importance in shaping fungi assemblages. Beyond the effect of plant distribution and

their interactions on the fungal assemblages, micro-scale changes in environment either or not

due to plant effect is likely to interplay with the recruitment and selection of microorganisms

isolates on plant roots. Further studies should explicitly quantify the effects of both local

535 environmental conditions and host-plant isolation, and apply the theoretical framework

developed for large spatial scales to the much smaller scale needed to examine assembly

patterns of microorganisms.

AcknowledgementsAcc

epte

d A

rtic

le

540 We would like to acknowledge A.-K. Bittebiere for plant maps, L. Uroy and A. Ernoult for

stimulating discussions on connectivity measurements, and M. Dahirel for advices on

statistics. This work has been funded by the EC2CO MICROLAND from CNRS.

References

545 Bahram, M., K.G. Peay, and L. Tedersoo 2015. Local-scale biogeography and spatiotemporal

variability in communities of mycorrhizal fungi. New Phytologist 205:1454-1463.

Barton K. 2016. MuMIn: Multi-Model Inference. R package version 1.15.6.

Bates, D., M. Maechler, B. Bolker, and S. Walker 2015. Fitting Linear Mixed-Effects Models

Using lme4. Journal of Statistical Software 67:1-48.

550 Bittebiere, A.K., P. Vandenkoornhuyse, E. Maluenda, A. Gareil, A. Dheilly, S. Coudouel, M.

Bahin, and C. Mony 2019. Plant spatial structure of plant communities determines arbuscular

mycorrhizal fungal community assembly. Journal of Ecology. https://doi.org/10.1111/1365-

2745.13279.

Bell, T. 2010. Experimental tests of the bacterial distance-decay relationship. ISME .J.

555 4:1357-1365.

Bennett, R., and B. Turgeon 2017. Fungal Sex: The Ascomycota. Pages 117-145 in: J.

Heitman, B. Howlett, P. Crous, E. Stukenbrock, T. James and N. Gow, editors The Fungal

Kingdom. ASM Press, Washington.

Bodenhausen, N., M. Bortfeld-Miller, M. Ackermann, and J.A. Vorholt 2014. A synthetic

560 community approach reveals plant genotypes affecting the phyllospere microbiota. PLoS

Genetics 10:e1004283.

Borneman, J., and R.J. Hartin 2000. PCR primers that amplify fungal rRNA genes from

environmental samples. Applied and Environmental Microbiology 66:4356-460.

Bulgarelli, D., M. Rott, K. Schlaeppi, E. Ver Loren van Themaat, N. Ahmadinejad, F.

565 Assenza, P. Rauf, B. Huettel, R. Reinhardt, E. Schmelzer, et al. 2012. Revealing structure and

assembly cues for Arabidopsis root-inhabiting bacterial microbiota. Nature 488:91–95

Bundrett, M., and B. Kendrick 1990. The roots and mycorrihzas of herbaceous woodland

plants. New Phytologist 114:457-468.

Calabrese, J.M., and W.F. Fagan 2004. A comparison-shopper’s guide to connectivity

570 metrics. Front Ecol Environ 2:529–536.Acc

epte

d A

rtic

le

Carvalho, L.M., P.M. Correia, R.J. Ryel, and M.A. Martins-Louçao 2003. Spatial variability

of arbuscular mycorrhizal fungal spores in two natural plant communities. Plant and Soil

251:227-236.

Carvalho, L.M., P.M. Correia, and M.A. Martins-Louçao 2004. Arbuscular mcorrhizal fungal

575 propagules in a salt marsh. Mycorrhiza 14: 165-170.

Chase, J.M. 2005. Towards a really unified theory for metacommunities. Functional Ecology

19:182-186.

Clapp, J.P., J.P.W. Young, J.W. Merryweather, and A.H. Fitter 1995. Diversity of fungal

symbionts in arbuscular mycorrhizas from a natural community. New Phytologist 130:259-

580 265.

Coleman-Derr, D., D. Desgarennes, C. Fonseca-Garcia, S. Gross, S. Clingenpeel, T. Woyke,

G. North, A. Visel, L.P. Partida-Martinez, and S.G. Tringe 2016. Plant compartment and

biogeography affect microbiome composition in cultivated and native Agave species. New

Phytologist 209:798-811.

585 Daniels, B.A., P.M. McCool, and J.A. Menge 1981. Comparative inoculum potential of

spores of six vesicular-arbuscular mycorrhizal fungi. New Phytologist 89:385-391.

Davison, J., M. Moora, T. Jairus, M. Vasar, M. Öpik, and M. Zobel 2016. Hierarchical

assembly rules in arbuscular mycorrhizal (AM) fungal communities. Soil Biology and

Biochemistry 97:63-70.

590 Des Abbayes, H., G. Claustres, R. Corillion, and P. Dupont 1971. Flore et végétation du

massif armoricain. I-Flore vasculaire. Pres. Univ. Bretagne. Saint Brieuc, France.

Duhamel, M., and P. Vandenkoornhuyse 2013. Sustainable agriculture: possible trajectories

from mutualistic symbiosis and plant neodomestication. Trends in Plant Science 18 :597-600.

Edwards, J., C. Johnson, C. Santos-Medellin, E. Lurie, N. Kumar Podishetty, S. Bhatnagar,

595 J.A. Eisen, and V. Sundaresan 2015. Structure, variation, and assembly of the root-associated

microbiomes of rice. Proceedings of the National Academy of Sciences USA, 112:E911-

E920.

Escudié F., L. Auer, M. Vernard, M. Mariadassou, L. Cauquil, K. Vidal, S. Maman, G.

Hernandez-Raquet, S. Combes, and G. Pascal 2017. FROGS: Find, Rapidly, OTUS with

600 Galaxy Solution. Bioinformatics 34:1287-1294.

Fagan, W., and J. Calabrese 2006. Quantifying connectivity: Balancing metric performance

with data requirements. In: Connectivity Conservation. Pages 297-317 in K. Crooks, and M.

Sanjayan, editors. Cambridge University Press, Cambridge.Acc

epte

d A

rtic

le

Fletcher, R.J., N.S. Burrell, B.E. Reichert, D. Vasudev, and J.D. Austin 2016. Divergent

605 perspectives on landscape connectivity reveal consistent effects from genes to community.

Current landscape ecology reports 1:67-79.

Fox, J., and S. Weisberg 2011. An {R} Companion to Applied Regression, Second Edition.

Thousand Oaks CA: Sage. URL: http://socserv.socsci.mcmaster.ca/jfox/Books/Companion.

Gollotte, A., D. van Tuinen, and D. Atkinson 2004. Diversity of arbuscular mycorrhizal fungi

610 colonizing roots of the grass species Agrosystem cappilaris and Lolium perenne in a field

experiment. Mycorrhiza 14:111-117.

Hanski, I. 1999. Connectivity, Habitat continuity, and metapopulations in dynamic

landscapes. Oikos 87:209-219.

Hanson, C.A., J.A. Fuhrman, MC. Horner-Devine, and J.B. Martiny 2012. Beyond

615 biogepgraphic patterns: processes shaping the microbial landscape. Nat. Rev. Microbiol. 10:

497-506.

Hart, M.M., and R.J. Reader 2002. Taxonomic basis for variation in the colonization strategy

of arbuscular mycorrhizal fungi. New Phytologist 153:335-344.

Johnson-Green, P.C., Kenkel, N.C. & Booth, T. (1995). The distribution and phenology or

620 arbuscular mycorrhizae along an inland salinity gradient. Canadian Journal of Botany, 73,

1318-1327.

Kiers, E.T., M. Duhamel, Y. Beesetty, J.A. Mensah, O. Franken, E. Verbruggen, C.R.

Felbaum, G.A. Kowalchuck, M.M. Hart, A. Bago, T.M. Palmer, S.A. West, P.

Vandenkoornhuyse, J. Jansa, and H. Bucking 2011. Reciprocal rewards stabilize cooperation

625 in the mycorrhizal symbiosis. Science 333:880-882.

König, S., T. Wubet, C.F. Dormann, S. Hempel, C. Renker, and F. Buscot 2010. TaqMan

Real-Time PCR assays to assess arbuscular mycorrhizal responses to field manipulation of

grassland biodiversity: effects of soil characteristics, plant species richness, and functional

traits. Applied and Environmental Microbiology 76:3765–3775.

630 Lê Van, A., A. Quaiser, M. Duhamel, S. Michon-Coudouel, A. Dufresne, and P.

Vandenkoornhuyse 2017. Ecophylogeney of the endospheric root fungal microbiome of co-

occurring Agrostis stolonifera. Peer Journal 5:e3454.

Lekberg, Y., and L.P. Waller 2016. What drives differences in arbuscular mycorrhizal fungal

communities among plant species? Fungal Ecology 24:135-138.

635 Leibold, M. A., M. Holyoak, N. Mouquet, P. Amarasekare, and et al. 2004. The

metacommunity concept: a framework for multi‐scale community ecology. Ecology Letters

7:601-613.

Acc

epte

d A

rtic

le

López-García, Á., C. Azcón-Aguilar, and J. Barea 2014. The interactions between plant life

form and fungal traits of arbuscular mycorrhizal fungi determine the symbiotic community.

640 Oecologia 176:1075-1086.

Lovett-Doust, L.L. 1981. Population dynamics and local specialization in a clonal perennial

(Ranunculus repens): I. The dynamics of ramets in contrasting habitats. Journal of Ecology

69:743-755.

Lundberg, D.S., S.L. Lebeis, S. Herrera Paredes, S. Yourstone, et al. 2012. Defining the core

645 Arabidopsis thaliana root microbiome. Nature 488:86-90.

Mahé, F., T. Rognes, C. Quince, C. de Vargas, and M. Dunthorn 2014. SWARM: robust and

fast clustering method for amplicon-based studies. Peer Journal 2:e593.

Martinez-Garcia, L.B., and F.I. Pugnaire 2011. Arbuscular mycorrhizal fungi host preference

and site effects in two plant species in a semiarid environment. Applied Soil Ecology 48:313-

650 317.

McArthur, H., and O. Wilson 1967. The Theory of Island Biogeography. Princeton University

Press, United States. NJ.

McGee, P.A., G.S. Pattinson, R.A. Heath, C.A. Newman, and S.J. Allen 1997. Survival of

propagules of arbuscular mycorrhizal fungi in soils in eastern Australia used to grow cotton.

655 New phytologist 135:773-780.

McMurdie, P.J., and S. Holmes 2013. phyloseq: An R Package for Reproducible Interactive

Analysis and Graphics of Microbiome Census Data. PLoS ONE 8:e61217.

McRae, B.H., and P. Beier 2007. Circuit theory predicts gene flow in plant and animal

poulations. Proceedings of the National Academy of Sciences USA 104:19885-19890.

660 McRae, B.H., B.G. Dickson, T.H. Keitt, and V.B. Shah 2008. Using circuit theory to model

connectivity in ecology, evolution, and conservation. Ecology 89:2712-2724.

Mony, C., P. Brunellière, N. Vannier, A.K. Bittebiere, and P. Vandenkoornhuyse 2019. Effect

of floristic composition and configuration on plant roots mycobiota: a landscape transposition

on a small scale. in press.

665 Nakagawa, S., and H. Schielzeth 2013. A general and simple method for obtaining R² from

generalized linear mixed-effects models. Methods in Ecology and Evolution 4:133-142.

Oksanen, J., F.G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P.R. Minchin,

R.B. O’Hara, G.L. Simpson, P. Solymos, M.H.H. Stevens, E. Szoecs, and H. Wagner 2018.

Vegan: community ecology package. – R package ver. 2.0-4, http://CRAN.R-project.org/

670 package vegan.Peay, K.G., T.D. Bruns, P.G. Kennedy, S.E. Bergemann, and M. Garbelotto Acc

epte

d A

rtic

le

2007. A strong species-area relationship for eukaryotic soil microbes: island size matters for

ectomycorrhizal fungi. Ecology Letters 10:470-480.

Peay, K.G., M. Garbelotto, and T.D. Bruns 2010. Evidence of dispersal limitation in soil

microorganisms: isolation reduces species richness on mycorrhizal tree islands. Ecology

675 91:3631-3640.

Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core Team 2018. nlme: Linear and

Nonlinear Mixed Effects Models. R package version 3.1-137.

Quast, C., E. Pruesse, P. Yilmaz, J. Gerken, T. Schweer, P. Yarza, J. Peplies, and F.O.

Glockner 2013. The SILVA ribosomal RNA gene database project: improved data processing

680 and web-based tools. Nucleic Acids Research 41:D590–D596.

Ries, L., R.J. Fletcher, J. Battin, and T.D. Sisk 2004. Ecological responses to habitat edges:

Mechanisms, models and variability explained. Annual Review in Ecology Evolution and

Systematics 35:491-522.

Scheublin, T., K. Ridgway, J.P. Young, and M.G.A. Van der Heijden 2004. Nonlegumes,

685 legumes and root nodules harbor different arbuscular mycorrhizal fungal communities.

Applied and Environmental Microbiology 70:6240-6246.

Shi, G., Y. Liu, L. Mao, S. Jiang, Q. Zhang, G. Cheng, L. An, G. Du, and H. Feng 2014.

Relative Importance of Deterministic and stochastic processes in driving arbuscular

mycorrhizal fungal assemblage during the spreading of a toxic plant. PLoS ONE 9:e95672.

690 Smith, S.E., and D.J. Read 2008. Mycorrhizal symbiosis. Third edition. Academic Press:

London.

Sullivan, L.L., B. Johnson, L.A. Brudvig, and N. Haddad 2011. Can dispersal mode predict

corridor effects on plant parasites ? Ecology 92:1559-1564.

Taylor, P.D., L. Fahrig, K. Henein, and G. Merriam 1993. Connectivity is a vital element of

695 landscape structure. Oikos 68:571–573.

Taylor, P.D., L. Fahrig, and K.A. With 2006. Landscape connectivity: a return to the basics.

In: Connectivity Conservation: maintaining connections for nature. Cambridge University

Press, Cambridge, pp 29–43.

Thrall, P.H., R. Godfree, and J.J. Burdon 2003. Influence of spatial structure on pathogen

700 colonization and extinction: a test using an experimental metapopulation. Plant Pathology

52:350-361.

Tischendorf, L., and L. Fahrig 2000. On the usage and measurement of landscape

connectivity. Oikos 90:7–19.Acc

epte

d A

rtic

le

Van der Heijden, M.G.A., J. Klironomos, M. Ursic, P. Moutoglis, R. Streitwolf-Engel, T.

705 Boller, A. Wiemken, and I.R. Sanders 1998. Mycorrhizal fungal diversity determines plant

biodiversity, ecosystem variability and productivity. Nature 396: 69-72.

Vandenkoornhuyse P., R. Husband, T.J. Daniell, I.J. Watson, J.M. Duck, A.H. Fitter, and

J.P.W. Young 2002. Arbuscular mycorrhizal community composition associated with two

plant species in a grassland ecosystem. Molecular Ecology 11:1555–1564

710 Vannier, N., C. Mony, A.-K. Bittebiere, S. Michon-Coudouel, M. Biget, and P.

Vandenkoornhuyse 2018. A microorganims’ journey between plant generations. Microbiome

6:79.

Warner, A., and B. Mosse 1980. Independent spread of vesicular-arbuscular fungi in soil.

Transactions of the British Mycological Society 74:407-410.

715 Wilson, D.S. 1992. Complex interactions in metacommunities, with implications for

biodiversity and higher levels of selection. Ecology 73:1984-2000.

Yang, H., Y. Zang, Y. Yuan, J. Tang, and X. Chen. 2012. Selectivity by host plants affects the

distribution of arbuscular mycorrhizal fungi: evidence from ITS rDNA sequence metadata.

BMC Evolutionary Biology 12:50–62.

720

Acc

epte

d A

rtic

le

Data Availability

Data are available on Figshare: https://doi.org/10.6084/m9.figshare.7893005.v1

725

Table. 1: Summary of the models explaining the fungal microbiota richness at the plot scale (i.e. based on the richness on roots of five B.

pinnatum plants sampled), which depended on mean isolation distance of host plants. Isolation was calculated as the mean pairwise distances

between all sampled host plants. Richness was log-transformed in order to normalize the distribution. The provided model parameters are the

intercept (Inter), independent variable with its coefficient (Est), the F statistic (F) and p-values from ANOVA and the adjusted R² (adj-R²).

730 Significant models are in bold. Resistance distance in 2011 and in 2012 are related with a correlation coefficient of 0.63 (Pearson correlation test,

t = 3.03; p<0.01).

Description Resistance dist. (2011) Resistance dist. (2012)

Mean (SD)

(min-max)

Inter Est F Adj-R² Inter Est F Adj-R²

All fungi 171.6 (23.9)

[131-206]

5.21 -0.003 2.15

(0.16)

0.07 5.20 -0.008 4.04

(0.06)

0.17

Ascomycota 116.4 (15.2)

[90-143]

4.78 -0.001 0.37

(0.55)

-0.04 4.80 -0.007 3.33

(0.09)

0.13

Basidiomycota 46.5 (14.6)

[20-72]

4.04 -0.1 4.73

(0.047)

0.20 3.92 -0.02 2.50

(0.14)

0.09

Glomeromycotina 8.8 (3.6)

[4-15]

1.80 0.01 3.55

(0.08)

0.15 1.91 0.02 2.76

(0.12)

0.11

Acc

epte

d A

rtic

le

Table. 2: Summary of the models explaining fungal microbiota richness and evenness on individual B. pinnatum plants, based on mean resistance

735 distance between B. pinnatum. Isolation was calculated as the mean pairwise distance to the closest sampled B. pinnatum. Richness was log-

transformed in order to normalize the distribution Given model parameters are the intercept (Inter), independent variable with its coefficient

(Est), the Chi-square statistic (Chisq) and p-values from ANOVA II and the marginal and complete R² (R²m; R²c). Significant models are in

bold. Resistance distance in 2011 and in 2012 are related with a correlation coefficient equal to 0.55 (Pearson correlation test, t = 5.48; p<0.001).

Description Resistance distance (2011) Resistance distance (2012)

Mean (SD)

(min-max)

Inter Est Chisq R²m(R²c) Inter Est Chisq R²m(R²c)

Richness

All fungi 70.0 (11.7)

[47-97]

4.21 0.001 0.71

(0.40)

0.015

(0.20)

4.24 0.003 0.008

(0.93)

0.0002

(0.21)

Ascomycota 47.6 (11.0)

[20-73]

3.78 0.003 1.61

(0.21)

0.029

(0.13)

3.84 0.0006 0.02

(0.88)

0.0004

(0.16)

Basidiomycota 18.0 (6.3)

[5-35]

2.89 -0.003 0.96

(0.33)

0.018

(0.14)

2.84 -0.002 0.17

(0.67)

0.003

(0.13)

Glomeromycotina 4.3 (2.3)

[1-13]

3.52 0.009 5.72

(0.02)

0.077

(0.079)

1.22 0.02 6.21

(0.01)

0.083

(0.08)

Evenness

All fungi 0.28 (0.06) 0.26 0.001 4.06 0.080 0.27 0.002 2.55 0.052 Acc

epte

d A

rtic

le

[0.14-0.36] (0.04) (0.28) (0.11) (0.32)

Ascomycota 0.35 (0.09)

[0.17-0.52]

0.32 0.001 2.37

(0.12)

0.052

(0.32)

0.34 0.002 2.34

(0.13)

0.049

(0.35)

Basidiomycota 0.32 (0.17)

[0.04-0.75]

0.26 0.003 4.64

(0.03)

0.063

(0.063)

0.31 0.0006 0.07

(0.80)

0.001

(0.04)

Glomeromycotina 0.38 (0.18)

[0.06-0.92]

0.46 -0.004 7.90

(0.005)

0.130

(0.22)

0.41 -0.004 2.60

(0.11)

0.049

(0.21)

740

Acc

epte

d A

rtic

le

Table. 3: Summary of the models explaining fungal microbiota Beta diversity (Bray-Curtis indices) on individual B. pinnatum plants, based mean

745 resistance distance between B. pinnatum. Isolation was calculated as the mean pairwise distance to the closest sampled B. pinnatum. Given model

parameters are the intercept (Inter), independent variable with its coefficient (Est), the Chi-square statistic (Chisq) and p-values from ANOVA II

and the marginal and complete R² (R²m; R²c). Significant models are in bold. Resistance distance in 2011 and in 2012 are related with a

correlation coefficient equal to 0.55 (Pearson correlation test, t = 5.48; p<0.001).

Description Euclidean distance Resistance distance (2011) Resistance distance (2012)

Inter Est Chisq R²m(R²c) Inter Est Chisq R²m(R²c) Inter Est Chisq R²m(R²c)

All fungi 0.51 (0.17)

[0.16-0.84]

0.55 -0.11 1.41

(0.23)

0.007 (0.57) 0.55 -0.002 3.92

(0.048)

0.053(0.56) 0.51 -0.001 0.21

(0.64)

0.003 (0.57)

Ascomycota 0.45 (0.17)

[0.12-0.77]

0.51 -0.16 3.64

(0.06)

0.018(0.48) 0.49 -0.003 5.35

(0.02)

0.068 (0.48) 0.45 -0.002 0.59

(0.44)

0.006(0.46)

Basidiomycota 0.63 (0.24)

[0.08-0.99]

0.69 -0.14 1.26

(0.26)

0.006(0.62) 0.69 -0.003 3.21

(0.07)

0.043(0.64) 0.62 0.002 0.29

(0.59)

0.003(0.61)

Glomeromycota 0.51 (0.24)

[0.05-0.96]

0.53 -0.05 0.09

(0.76)

0.007(0.18) 0.53 -0.0009 0.37

(0.54)

0.004(0.18) 0.48 0.005 3.16

(0.08)

0.030(0.21)

Acc

epte

d A

rtic

le

750 Figure legends

Figure 1 - Distribution of fungal microbiota assemblages for Ascomycota, Basidiomycota, and

Glomeromycotina based on OTU composition recorded in total. Scales on the Y axis differ by

fungal groups: Ascomycota, Basidiomycota and Glomeromycotina represented 71.6%, 27.8% and

0.6% of the sequences respectively. Left graphs depict the percentage of sequences and the right

755 graphs depict OTU richness for the main taxonomic groups represented within Ascomycota,

Basidiomycota, and Glomeromycotina.

Figure 2 - Mean isolation distances at the plot scale for the four mixtures of plants we analyzed.

1sp. denotes the B. pinnatum monocultures, whereas the other ‘sp.’ abbreviations denote B.

pinnatum in mixed plant mesocosms of 2, 4, and 8 plant species. Bar graphs provide isolation

760 distances, Euclidean distances (green), Resistance distances calculated in 2011 (yellow) and in

2012 (red).

Acc

epte

d A

rtic

le

ecy_2976_f1.tif

Acc

epte

d A

rtic

le

ecy_2976_f2.tif

Acc

epte

d A

rtic

le